Professional Documents
Culture Documents
19990315 084
Approved for public release; distribution is unlimited.
The findings in this report are not to be construed as an official
Department of the Army position unless so designated by other
authorized documents.
u
Table of Contents
Page
1. Introduction 1
2. The Momentum Equation and Special Cases 2
3. Lagrangian vs. Eulerian Acceleration 4
4. Noninertial Reference Frames 5
5. Extended Bernoulli Equation 8
6. Nonsteady Potential Flow Around a Sphere 9
7. Nonsteady Solid Eroding-Rod Penetration 12
8. Consequences of Noninertiality 19
9. Conclusions 21
10. References 25
Distribution List 27
Report Documentation Page 43
in
INTENTIONALLY LEFT BLANK.
IV
1. Introduction
The Bernoulli equation is perhaps the most famous and widely used equation in fluid
mechanics, relating the pressure, p, in a flow to the local velocity, V, and gravity potential. It
was derived by considering a balance of momentum along a streamline, for the special case of
steady, incompressible, inviscid flow in an inertial reference frame, with gravity as the only
significant body force. The Bernoulli equation may also be derived from considerations of
energy conservation, since, for inviscid flows, there is no energy loss. The Bernoulli equation
is given as
V2
p— + p + pgh = constant , (1)
where p is the flow density, g is the acceleration due to gravity, and h is the vertical height of
the flow relative to some reference location. Furthermore, if the flow is irrotational, the constant
of eqn (1) will be the same for all streamlines throughout the flow. The real world is rarely so
kind as to satisfy all the restrictive conditions under which eqn(l) was derived. Yet, because
the influence of these nonideal (compressible, viscous, rotational, accelerational) terms is often
small, the engineering world makes great use of eqn (1), often modifying it in an ad hoc manner
when nonideal effects rear their ugly head.
We endeavor here to pull together various equations and constructs from the literature into
a single framework, to present an unsteady, compressible, rotational, elasto-viscoplastic,
noninertially referenced momentum equation with no presuppositions. The importance of each
term can then be examined at the time of application to ascertain when discarding or
approximating it is appropriate. Because our primary interest in the subject lies in the area of
noninertial coordinate systems, examples of this variety, which make use of what might be called
an extended Bernoulli equation, are presented.
All of the concepts relating to the momentum equation that are discussed in this report
are readily available in one form or another throughout the educational literature of fluid
mechanics, solid mechanics, and dynamics. They are, however, not always found in a single
1
location. Furthermore, in an effort to teach textbook examples and solve textbook problems, the
educational literature quickly reduces the governing equation to certain well-known academic
cases, often failing to give full coverage to the general case involving viscous (or rotational),
compressible, accelerating, nonsteady flows in noninertial coordinate systems.
For example, Shames [1] generally does an excellent job of covering most aspects of the
momentum equation and noninertial reference frames, though it is done in terms of finite-sized
control volumes and not streamline-sized "flow tubes." Potter and Foss [2] cover all the relevant
equations regarding the forces and accelerations upon a material point in a flow, but fail to tie
the equations together into a generalized unsteady Bernoulli equation. Kelley [3] derives an
extended Bernoulli equation, but only for the case of nonviscous flows in inertial coordinate
systems. Currie [4] also derives a restrictive form of the extended Bernoulli equation, valid only
for irrotational flow in an inertial reference frame. The very thorough Schlichting [5], because
of its emphasis on boundary layers, does not even address the issue of noninertial coordinate
systems. Greenspan [6] examines the momentum equation in a noninertial frame, but only for
the special case of purely rotating frames, as might be found in the case of rotating fluid
problems. In addition to addressing the steady-state Bernoulli equation for streamlines, Lamb [7],
like Shames [1], also covers aspects of noninertial frames, but on an integrated volume basis.
Thus, this report is intended merely to serve as a handy repository of several important
well-established concepts that might otherwise need to be tracked down in a multiplicity of texts
and chapters.
OV s... - Vp „
_ = -±L L + VO , (2)
Dt p
where DIDt denotes the material derivative (discussed in following section); V is the vector
velocity of the material element, as measured in an inertial reference frame; p is the element
pressure; p is the element density; <£ is the body force potential; V is the vector gradient
operator; s(J is the deviatoric-stress tensor arising from any type of elasto-viscoplastic constitutive
behavior; and s^j is index notation for dsy/dxp denoting the following vector condensation of the
deviatoric-stress tensor:
ds ds ds * ds ds ds * ds ds ds
s... = (—- + -J2L + _Ji) i + (-Jl + -JL + _£.); + (_ü + __£ + _ü) k . (3)
'■''■' dx dy dz dx dy dz dx dy dz
Eqn (2) is a general form of the momentum equation from which many commonly employed
special cases derive.
For example, when deviatoric stresses, sip are zero, as in the case of an inviscid fluid, the
momentum equation, eqn (2), becomes the well-known Euler's equation,
- --JL + VO . (4)
Dt p
For a body in equilibrium, where the material accleration, DVIDt, is everywhere zero, but where
deviatoric stresses, stj, may arise from elastic strains in the body, eqn (2) reduces to the
equilibrium equation of solid mechanics,
where aijwj is the absolute-stress tensor condensation resulting from the combination of the
pressure gradient and deviatoric-stress condensation. On the other hand, if the flow is
accelerational, but the deviatoric stresses in eqn (2) arise solely from Newtonian viscosity, u, in
which shear stress is proportional to the associated velocity gradients (and assuming the validity
of Stokes' hypothesis), then the deviatoric-stress condensation can be expressed in terms of
velocity gradients (e.g., Schlichting [5]) to give the famous Navier-Stokes equation,
For incompressible viscous flow, the last term of eqn (6) will vanish, since, for incompressible
flow, the divergence of velocity is identically zero. This incompressible form of eqn (6) is known
as the incompressible Navier-Stokes equation. Eqns (4)-(6) each represent a useful special-case
solution of the general momentum equation, eqn (2).
4
-^V[x(t),y(t),z(t),t] = BY. = ^ + (V-V)V . (7)
rf? Dt dt
In addition to the local acceleration, it is seen that the material acceleration is composed of terms
known as acceleration of transport, or convective acceleration, given by the last term of eqn (7).
This equation reveals how conditions in a flow field at all points fixed in space can be steady
(dV/dt = 0), while, at the same time, any material element of that flow experiences all manner of
accelerations as it traverses the field (DV/Dt^O).
Furthermore, a number of texts (e.g., Potter and Foss [2]) also present a form of eqn (7)
that has been manipulated via vector mechanics, to yield
2
DV = —
dV + yyfV
V ^
+ (VxV)xV . (8)
Dt dt 2
v ,
This form is especially interesting because it separates the V2 inertial-force term from the
vorticity-induced term involving cross products. For flows that are irrotational, all terms
involving vorticity, V x V, will vanish. Furthermore, the inertial-force term is the genesis of the
V2 dependence of the Bernoulli equation, eqn (1).
Any undergraduate dynamics text (e.g., Beer and Johnson [9]) and many fluid mechanics
texts (e.g., Shames [1] and Potter and Foss [2]) derive or present the equation for acceleration of
a particle, when the kinematics of particle motion are measured with respect to a noninertial
reference frame xyz. Using the notation of Figure 1, in which the noninertial frame xyz moves
with respect to an inertial reference frame XYZ by way of translation vector S(t) and rotation
vector Q(0, while the kinematics of the particle motion in question are measured with respect
to xyz by the displacement vector R(t), one obtains
where A is the total acceleration with respect to XYZ; Vm = dRIdt is the velocity measured in the
noninertial xyz frame; and a =DVxyJDt is the material acceleration, as measured in xyz.
Since eqn (2), the momentum equation, can only be valid when applied in an inertial
frame, eqn (9) provides the means to apply eqn (2) with respect to the inertial XYZ, even when
the flow kinematics (e.g., R and V) are measured with respect to a translating and rotating xyz,
which are perhaps attached to a body of interest. Realizing that the inertial acceleration, A, in
eqn (9) corresponds to the material acceleration, DVlDt, presented on the left side of the inertially
constrained eqn (2), we have by substitution
S
A = —ZL + — + 2QxV + QxjQxR) + ^IxR = ^~Vp + V<D , (10)
Dt dt2 dt p
Substitution of eqn (8) for the noninertial xyz material acceleration, DVmIDt, and some simple
rearrangement gives the following result:
v
m + <£S_ + (VxV
v )xy + 2QxV + Qx(£lxR) + —xR
v
,.•> xyz' xyz xyz ' j
dt dt2 m
* "* dt t
s...
,J,J
- Vp W*
= V$ + - — (ID
P 2
Eqn (11) is valid at all points in a compressible, elasto-viscoplastic, rotational, nonsteady flow
subject to conservative body forces, as measured in a reference frame undergoing time-dependent
translational and rotational motions.
Figure 1. Depiction of the noninertial reference frame xyz translating (S) and rotating (Q)
with respect to inertial frame XYZ. Flow kinematic variables R, Vxy7? and a are
measured with respect to the noninertial xyz frame.
The first two terms of eqn(ll) represent the inertial XYZ rigid-body translation^
acceleration of the material point in question. The third term, involving vorticity, V x V,
represents acceleration resulting from flow vorticity, as measured in xyz. The remaining terms
on the left-hand side represent Coriolis, centripetal, and rotational accelerations arising strictly
from the time-dependent rotational motion of the noninertial reference frame xyz. On the
right side of eqn(ll), the traditional Bernoulli force terms (body, pressure, and inertial) as well
those involving deviatoric-stress gradients are found.
where the vector increment dR is made to follow the path of the contour throughout the
integration. This result is valid for nonsteady, compressible, rotational, elasto-viscoplastic flows
in a noninertial reference frame. Note that a minor vector manipulation has been performed upon
the vorticity integral term. Furthermore, the gradient integrals of inertial and body forces on the
right-hand side of eqn (12) were reduced to a difference in the values of V2/2 and <3> between the
two endpoints of the contour. The pressure gradient integral may also be a function of the
contour endpoint values, {pip), but only if the flow is incompressible; otherwise, the term must
be integrated along the contour. Unfortunately, the deviatoric-stress integral must, in general, be
8
explicitly performed, as it does not represent a gradient potential. The contour of integration in
eqn(12) may be any arbitrary three-dimensional contour. However, the resulting equation,
because of the vector mathematics, will be scalar.
Different types of problems will employ different terms of this equation. For example,
problems of fluids involving turbomachinery will allow the first integral to be discarded if the
problem, when viewed in a rotating coordinate system can be made to appear steady. For
irrotational flows, such as are found in many applications of potential flow theory, the second
integral may be discarded. Even when the flow is rotational, if the integration contour is, at a
particular instant in time, also a streamline (i.e., everywhere parallel to the velocity vector), the
second integral also disappears, as VxdR will be zero at all places along a streamline. For the
problem of linearly accelerating bodies within a flow field, the noninertial body-coordinate
system xyz need not rotate and the third integral may therefore be discarded for problems of this
type. Typical conditions that could justify elimination of terms on the right-hand side of the
equation would involve negligible body forces and/or shear stresses (inviscid, nonelastic).
Several examples involving the use of eqn(12) are now investigated. Because our
primary interest in the subject lies in the area of noninertial coordinate systems, we will focus
on problems of this type.
Fortunately, eqn (12) allows one to overcome this difficulty. An unsteady potential flow
problem, in which the universe is allowed to accelerate about a fixed-in-space potential-flow
body, can be solved, as long as it is realized that the potential-flow body is fixed in noninertial
xyz space—a coordinate system that is, in effect, accelerating equal and opposite to the
acceleration of the potential-flow far-field. In this way, the net far-field acceleration is zero and
the unsteady motion of a body through an inertial flow field is truly modeled. This argument
is valid for both linear and rotational far-field accelerations.
where U represents the uniform free-flow velocity about a sphere of radius r0, fixed at the origin
of the coordinate system (with the flow traveling from the -x toward the +x direction). To make
this flow unsteady, allow the free-flow velocity to be a function of time, U(t). Recall, to avoid
the complication of trying to force the universe to accelerate around the sphere, that the potential-
flow coordinate axes, xyz, attached to the sphere, are in fact simultaneously traveling toward the
-x direction with a nonsteady velocity of magnitude U(t).
Realize that this problem does not involve vorticity, employs a noninertial reference frame
that does not rotate, has negligible body forces, has no shear stresses (inviscid), and is
incompressible. The extended Bernoulli equation for this problem then becomes
10
^V„
xyz d*S\ ■dR
Jn [Vlv (14)
f dt
+
dt i 2
= -x
2 p
where /?, and R2 are the endpoints of the integration contour. If the integration contour is chosen
to be the (straight line) stagnation contour traversing from (-°°,0,0) to (-r0,0,0), the only velocity
component of relevance to the integral is the x component, so that
VJiX'V = -vJe=* = u
® C1 + r
o^3) (for x<-r0,y = z = 0) . (15)
The terms from the right-hand side of eqn(14) may thus be evaluated as follows:
l
(16)
i\dt dt) p 2 pI
To finish the solution, dVJdt needs to be evaluated from eqn(15) and substituted into
eqn (16). Time-dependent potential flows (and others) often have the virtue of being separable
in space and time, as in Vx(x,t) = U(t)-g(x). If the one-dimensional (1-D) contour length is
infinite in extent, the spatial integral of the time derivative, in this separable case, may be
expressed as
]!Lfc.££-]2-*.
J dt dt 1 U
(i7)
-oo -oo
Alternately, if the 1-D contour were of finite length, eqn (17) could be expressed as
\^ldx-^L(b-a)-A (18)
J dt dt Uib-a)
The right side of eqn (18) is composed of the free-stream acceleration multiplied by the contour
length as well as a quotient factor. The quotient factor is the average velocity along the contour
divided by the free-stream velocity, and, in the case of a stagnation contour, it will generally fall
in the range from zero to unity, depending on the details of the flow.
11
In the present case, eqn(17) is utilized for the evaluation of eqn(16). It is noted that
there is a canceling of the dUldt term, which is necessary to avoid computing the force necessary
to accelerate the universe. For the considered flow integrated along the specified contour,
eqn (16) may be evaluated as
pU2 + Pr dU
p -p = —— —0 . nm
(19)
Thus, if the time-dependent velocity of the sphere is known, the stagnation pressure may be
evaluated with eqn (19). That pressure varies from what would be predicted by the Bernoulli
equation, eqn (1), by a term involving the acceleration rate of the sphere. If the acceleration of
the sphere is positive, the stagnation pressure is seen to be higher than the Bernoulli pressure,
while, if the sphere is decelerating, the stagnation pressure is less.
A more recent treatise on the subject, which instead relies upon a force/momentum
balance along the centerline contour only, is that of Walker and Anderson [15]. Upon assuming
certain reasonable velocity fields in the tip of the rod and in the target crater, they proceed to
solve the momentum equation in glorious detail, directly in the inertial XYZ laboratory frame of
12
reference, to include noninertial effects. The analysis presented here is not intended to supplant
the esteemed work of Walker and Anderson. Rather, it is intended to show that the concepts
derived herein may be very simply applied to the same problem to a similar end. Furthermore,
the manner in which an accelerating coordinate system, attached to the rod/target interface,
affects the overall result should be apparent in a more direct way.
In the eroding-rod problem (see Figure 2), a solid rod, of density pR, instantaneous length
L, and velocity V, penetrates into a semi-infinite block of density pT. The rod is assumed to
support a uniaxial-stress state in the longitudinal direction of the rod. The eroding interface is
traveling into the target at velocity U. Furthermore, employing the assumed velocity profiles
suggested by Walker and Anderson, there is a small plastically deforming region located at the
eroding tip of the rod, of length s, where the velocity linearly transitions from the rigid-body rod
velocity of V to the interface velocity of U. On the target side of the interface, the crater
geometry (Figure 3) is locally considered a hemisphere of radius R in polar (r,0) coordinates,
with the target flow velocity, u, along the axis of symmetry decaying as
2
u 1 aR -1 (R<r<aR) , (20)
2
U a -1 L
while remaining zero at all distances r at and beyond aR. The parameter a defines an extent of
plasticity in the target, with a> 1 defining a finite-sized plastic zone in the target, and a—»1*
denoting the limiting case of infmitesimally thin plastic zone. Along the axis of symmetry of
the noninertial xyz coordinate system of Figure 3, z = r - R. Both the projectile
«
and target plastic-
flow zones may be considered incompressible, despite the axial velocity gradients, because of an
associated radial divergence of the flow field.
Because V, U, and L are changing with time t, the reference frame attached to the
rod/target interface will be a noninertial frame xyz traveling at the time-dependent velocity U.
Though Walker and Anderson considered the general case of time-varying s and a, these
geometry parameters are held constant for simplicity. From the perspective of the interface
13
Plastic Zone Plastic Zone
in Rod in Target
Figure 3. Assumed target flow pattern in target, per Walker and Anderson [15]. Note that,
along the axis of symmetry, the crater coordinate, r, is related to interface
coordinate, z, by z = r - R.
14
coordinate system xyz, the velocity as a function of axial coordinate z, along the centerline of the
problem, can be given as
v-u -L<z< -s
-(V - U) zls -s<z<0
Vz = U -U 0<z<(a-l)R
a" 1 R+z
-U z>(a-l)R (21)
This flow field is schematically shown in Figure 4. Specifying the extended Bernoulli equation,
eqn (12), along the contour defined by the axis of symmetry, for this case of irrotational (along
the axis of symmetry), incompressible flow with only rectilinear accelerations, the extended
Bernoulli equation reduces into a 1-D integration in z, yielding
2 Z, *2
V. 2oXZ,X +GZZ,Z \J
,
'♦£!W--.ü
dt I ~2
+ (22)
fi z, z
The factor of 2 on the axlJ, term arises because of symmetry, for which oxv and ayzy are equal
on the axis of symmetry. Furthermore, because the integration contour is aligned with the z axis
and the flow is incompressible, the G^ integral will amount to a difference of (o^/p) between
the contour endpoints. This equation is, of course, identical to the momentum equation derived
by Walker and Anderson, though expressed in the noninertial xyz coordinates, rather than the
laboratory XYZ coordinate frame.
First, limit the fixed integration contour to the elastic portion of the solid rod, spanning
the range -L<z< s, and solve eqn(22) in light of the velocity field specified by eqn(21).
Because the rod velocity at z = s and z = -L are identical, the contribution of the V2 gradient
integral is zero. Further, because the stress state in the rod is assumed uniaxial in z, the shear-
stress-gradient integral will be exactly zero. Finally, note how the dUldt acceleration terms from
V, and the noninertial frame acceleration cancel out. Thus, one obtains
15
Figure 4. Schematic depicting the assumed velocity field along the axis of symmetry of the
eroding rod and target.
16
ÄL-5) = _L(-7Ä-0) , (23)
dt PÄ
where YR is the yield strength of the rod, being exactly the uniaxial-stress state (where tension
is defined as positive) at the elastic-plastic interface [i.e., aa(z = -s) = -YR]. This is the
high-sound-speed limiting result of Walker and Anderson (since it was not here accounted for
the finite wave speed at which acceleration information travels down an elastic bar). They noted
further that, were the size of the rod's plastic zone, s, a negligible percentage of the overall rod
length, L, eqn (23) reduces to the Alekseevskii-Tate rod deceleration equation, dVldt = -YR/(pRL).
Despite any decelerations of the noninertial frame traveling at velocity U with the rod/target
interface, the rod deceleration equation is totally independent of interface velocity U.
Secondly, reconsider eqn (22) over a different integration contour, still along the axis of
symmetry but spanning from -L<z<0, thereby including the complete rod in the integration.
The terms of eqn (23) are thus retained, while adding to them the terms that arise from the small
plastic zone at the tip of the eroding rod. Denoting the axial stress aa, at the rod/target interface,
as ostag, one obtains
This result is identical to the result of Walker and Anderson, for the case of constant plastic zone
extent, s. It can be solved for the compressive stagnation stress at the rod/target interface and,
by making use of a substitution of eqn (23), results in
V Uf
-o....-^
stag : ^Y.-i^^-C.
^dt + dt'
(25)
If the extent of the rod's plastic zone, s, is small compared to rod length, L, or if the penetration
process is steady (i.e., velocity derivatives zero), the last term in eqn (25) becomes negligible and
the remaining terms become identical to the expression proposed by Täte [11] for the stagnation
stress on the rod side of the rod/target interface.
17
Turning to the target, the integration contour is defined to be along the axis of symmetry
throughout the target. That is, eqn (22) is integrated between 0 < z < °°, in light of eqn (21), to
obtain
(a 1)Ä
dUldt (aRy^2 +z ' f dU + dU]
cc2-l (R + z) Jo dt dt J (a-l)S
(o-l)fi
dx dx (26)
2 PT PT i Pr Jw •
The last term of both sides of the equation (the rigid-body acceleration term on the left, and the
shear-stress integration on the right) are both zero, since the target material beyond the region
of plastic extent, z > (a - l)R, is essentially undisturbed. Solving for the axial stagnation stress
on the target side of the rod/target interface gives
Since this paper is primarily concerned with the kinematics of nonsteady flow fields, it
is not intended to delve into the constitutive relations by which the shear-stress integral along the
centerline contributes to the stagnation pressure beneath an eroding rod. Walker and
Anderson [15] may be consulted for these details for those interested. Suffice it to say that the
terms in eqn (27) correspond identically to their terms associated with target stresses, for the
special case of fixed extent of target plasticity (i.e., constant a). Furthermore, they note that, for
the limiting case of small crater radius, R -> 0 (corresponding to truly 1-D penetration), the shear-
stress integral becomes the sole modification to the Bernoulli stagnation pressure. For the
fixed-a case, this shear-stress integral becomes a positive constant related to the yield strength
of the target material and is traditionally given the name target resistance, denoted Rr One may
infer from the result of Walker and Anderson that, in addition to the target's inertial head
(pr£/2/2), it is the target's shear-stress field, rather than the acceleration of target material under
the penetrator, that is the primary contributor to interface pressure on the target side of the
interface, when the penetration process is nearly steady.
18
By limiting the scope of problem complexity and by achieving algebraic expediency via
the use of a noninertial rod/target interface coordinate system, the primary result of Walker and
Anderson, who spent quite a number of journal pages exhaustively addressing this subject, has
been recreated in the span of several paragraphs. For those who don't wish to dwell on the solid-
mechanics aspects of their derivations, the parts of the problem dealing with accelerations and
noninertial frames can be grasped here, in their essence.
8. Consequences of Noninertiality
The means of accounting for the noninertiality of a reference system have long been
established and are embodied in eqn(9). Failure to properly take these terms into account,
however, will lead to erroneous calculations in various forms. Consider the two example
problems examined in the proceeding text to see the consequences of improperly applying the
momentum equation in a noninertial frame.
For the accelerating sphere problem, failing to subtract out the dUldt acceleration of the
noninertial xyz frame would have added a term to the stagnation pressure, eqn (19), of magnitude
p • dU/dt multiplied by the contour length, call it I. Obviously, for a contour of infinite length,
the error would be infinite, resulting from the fact that the pressure being computed arose from
accelerating the whole mass of the universe about the sphere. If the contour length were finite,
the added pressure term, being proportional to contour length, is like the situation existing within
a (inviscid) wind tunnel. Additional pressure head needs to be supplied to the tunnel in order
to accelerate the flow through the tunnel test section. A quick inspection of the form of eqn (19)
(augmented on the right side by p / • dUldt) reveals that, as the pressure differential is raised
across the test section, the flow velocity will accelerate to eventually reach a new equilibrium
velocity. The length of the test-section contour, /, denoting the length (and thus mass) of the
flow to be accelerated, will govern the time constant of the acceleration. So, in the case of the
problem of an accelerating sphere, improperly ignoring the noninertiality of the reference frame
actually changes the problem to one of a sphere fixed in a wind tunnel.
19
For the eroding-rod problem, the consequences of ignoring the noninertiality of the
interface coordinate system produce a different set of errors. The consequences upon the rod
deceleration equation, eqn (23), would literally be to replace dVldt with d(V - U)/dt, as in
^-ü)(r-.)B-y« (28)
dt 9R
For a symmetric impact of like materials at speeds above the elastic limit, U will typically be on
the order of V72. In such a case, the effect on rod deceleration will be an error on the order
of 100%. For cases where U is a larger percentage of V, as in the case of high-density-rod
penetration, the error in the deceleration calculation is correspondingly increased.
On the target side of the interface, failure to account for the acceleration of the coordinate
system will introduce a -pT-l-dU/dt contribution to right side of the momentum balance
equation, eqn (27), as in
-o, = ^ + TR£L^1
PT - 2°T^L& - pTl™ . (29)
'"" 2 dt a+1 I Bx dt
Here, / denotes a contour length of target material to be integrated [assumed greater than or equal
to (a - 1)R], and dUldt is negative for a decelerating rod. The first warning flag is that the last
term of eqn (29) is proportional to the contour length which, for a semi-infinite target, is infinite
in length. Such an improper inertial interpretation again leads to a calculation of the force to
accelerate the universe with respect to the rod/target interface. If, on the other hand, the
thickness of the target were finite, or if the integrated contour length, /, were arbitrarily made
finite, eqn (29) though quite incorrect, might seem less obviously so. If length / of the integrated
contour were large enough to dominate the other terms of eqn (29), leading to
a,stag =pTT/—
dt
, (30)
one might erroneously conclude that the normal interface stress, oslag, is primarily supported by
the "apparent" deceleration of target material relative to the rod/target interface, rather than by
20
the inertial head and elastic shear-stress distribution within the target. In reality, the true effect
of interface deceleration [second term on the right side of eqns (27) and (29)] has just the
opposite effect (i.e., opposite sign): when the interface and associated target material are
traveling at velocity U, a deceleration of the interface actually lowers the stagnation stress
because not only is U made lower in the process, but also the associated target material is
inertially tending to travel at U and resists any decrease in interface velocity. This resistance of
target material to decelerate (i.e., the inertia of the plastically entrained crater material) would
have the effect of superimposing an axial-tension field on top of the steady-state (inertial)
compression distribution. Thus, the act of interface deceleration actually lowers the interface
stress.
Another reality check, which would indicate the inappropriateness of eqn (30), is the
inference that a positive acceleration of the rod/target interface would be met with tension at the
interface. Such accelerations invariably occur, when the penetration of a multilayered target
transitions from a high-density target element to a lower density element of comparable strength.
Yet, it is known that such a transition is not accompanied by tension at the rod/target interface.
Thus, in the case of an eroding rod undergoing deceleration, it may be concluded that a proper
accounting of the noninertial behavior of the rod/target interface is crucial to a proper formulation
of the overall problem.
9. Conclusions
Once the groundwork for the extended Bernoulli equation, eqn (12), has been laid, the
solution to actual problems can often proceed quickly. All of the concepts necessary to develop
this equation have existed in the educational literature of fluid and solid mechanics and dynamics
for many years, if not centuries. However, all of the applicable terms contributing to the
equation are not generally located in a single source, as the educational literature prefers to
expeditiously reduce the governing momentum equation to special-case solutions for instructional
purposes. These special-case limitations often include steady, incompressible, irrotational, or
inviscid flows in inertial reference frames.
21
The momentum equation along an integration contour within a general flow field has been
herein rederived. By placing no restrictions on the type or manner of flow, the equation has been
presented, using the popular terminology, as an "extended" Bernoulli equation. The equation,
as presented, is valid for unsteady, compressible, rotational, elasto-viscoplastic flows measured
relative to a noninertial (translating and/or rotating) coordinate system, whose motion is known.
In the case of the solid eroding-rod problem, by comparing the present analysis to that
of Walker and Anderson [15] (who solved the identical problem in the inertial laboratory frame
of reference), it was observed that choosing a convenient coordinate system (even if noninertial)
can significantly simplify the algebraic manipulation of the governing momentum equation.
However, if misapplied, the consequence of failing to account for the acceleration of the eroding
interface produces significant errors, numerically and conceptually. First, the rod deceleration
rate is miscalculated, often by a factor of 2 or greater. Also, in the target, the basic
understanding of the problem is completely distorted by failure to properly account for the
noninertiality of the interface reference frame. In reality, the inertial head and elastic shear-stress
distribution within the target are primarily responsible for the buildup of interface pressure, while
the interface deceleration, because of the target-material inertia in the plastically entrained zone
of the target, actually ameliorates the interface stress. From the point of view of the noninertial
frame however, one might erroneously conclude that the interface deceleration was actually the
primary cause for the buildup of stress on the target side of the interface—a conclusion totally
opposite from and in contradiction to the properly formulated (inertial) momentum balance.
22
This report presents a general form of the momentum equation that is extremely useful
for solving a great variety of problems that might not otherwise fall into idealized categories.
The solved examples help to illustrate the power of choosing a convenient frame of reference in
which to solve a given problem. However, the examples also serve to emphasize the vital
importance of properly accounting for effects of accelerating reference frames.
23
INTENTIONALLY LEFT BLANK.
24
10. References
1. Shames, I. H. Mechanics of Fluids. McGraw-Hill: New York, 1962.
3. Kelley, J.B. "The Extended Bernoulli Equation." Amer. J. Phys., 18, pp. 202-204, 1950.
5. Schlichting, H. Boundary Layer Theory. Seventh (English) Edition, McGraw-Hill: New York,
1979.
9. Beer, F. P. and E. R. Johnson, Jr. Vector Mechanics for Engineers: Dynamics. Third Edition,
McGraw-Hill: New York, 1977.
10. Alekseevskii, V.P. "Penetration of a Rod into a Target at High Velocity." Comb. Expl. and
Shock Waves, 2, pp. 63-66, 1966.
11. Täte, A. "A Theory for the Deceleration of Long Rods After Impact." J. Mech. Phys.
Solids, 15, pp. 387-399, 1967.
12. Täte, A. "Long Rod Penetration Models—Parti. A Row Field Model for High Speed Long
Rod Penetration." Int. J. Mech. Sei., 28: 8, pp. 535-548, 1986.
13. Täte, A. "Long Rod Penetration Models—Part II. Extensions to the Hydrodynamic Theory
of Penetration." Int. J. Mech. Sei., 28: 9, pp. 599-612, 1986.
15. Walker, J. D. and C.E.Anderson, Jr. "A Time-Dependent Model for Long-Rod
Penetration." Int. J. Impact Engng., 16:1, pp. 19-48, 1995.
25
INTENTIONALLY LEFT BLANK.
26
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
HQDA DIRECTOR
DAMOFDQ US ARMY RESEARCH LAB
D SCHMIDT AMSRLDD
400 ARMY PENTAGON JJROCCHIO
WASHINGTON DC 20310-0460 2800 POWDER MILL RD
ADELPHI MD 20783-1145
OSD
OUSD(A&T)/ODDDR&E(R) DIRECTOR
RJTREW US ARMY RESEARCH LAB
THE PENTAGON AMSRL CS AS (RECORDS MGMT)
WASHINGTON DC 20301-7100 2800 POWDER MILL RD
ADELPHI MD 20783-1145
DPTY CG FOR RDE HQ
US ARMY MATERIEL CMD DIRECTOR
AMCRD US ARMY RESEARCH LAB
MG CALDWELL AMSRL CILL
5001 EISENHOWER AVE 2800 POWDER MILL RD
ALEXANDRIA VA 22333-0001 ADELPHI MD 20783-1145
DARPA
B KASPAR
3701 N FAIRFAX DR
ARLINGTON VA 22203-1714
US MILITARY ACADEMY
MATH SCI CTR OF EXCELLENCE
DEPT OF MATHEMATICAL SCI
MAJMDPHTTIJPS
THAYERHALL
WEST POINT NY 10996-1786
27
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
COMMANDER 4 COMMANDER
US ARMY ARDEC NAVAL WEAPONS CENTER
AMSTA AR CCH V M D NICOLICH N FASIG CODE 3261
PICATINNY ARSENAL NJ 07806-5000 T T YEE CODE 3263
D THOMPSON CODE 3268
COMMANDER W J MCCARTER CODE 6214
US ARMY ARDEC CHINA LAKE CA 93555
E ANDRICOPOULOS
PICATINNY ARSENAL NJ 07806-5000 12 COMMANDER
NAVAL SURFACE WARFARE CTR
COMMANDER DAHLGREN DIVISION
USA STRATEGIC DEFNS CMD HCHEN
CSSD H LL T CROWLES D L DICKINSON CODE G24
HUNTSVELLE AL 35807-3801 C R ELLINGTON
C R GARRETT CODE G22
COMMANDER W HOLT CODE G22
US ARMY MICOM W E HOYE CODE G22
AMSMI RD ST WF R MCKEOWN
SHELL J M NELSON
D LOVELACE MJ SELL CODE HI 1
M SCHEXNAYDER W J STROTHER
REDSTONE ARSENAL AL 35898-5250 A B WARDLAW JR
L F WILLIAMS CODE G33
MIS DEFNS & SPACE TECHNOLOGY 17320 DAHLGREN RD
CSSD SD T K H JORDAN DAHLGREN VA 22448
PO BOX 1500
HUNTSVILLE AL 34807-3801
28
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
29
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
30
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
31
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
32
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
33
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
34
NO. OF
COPIES ORGANIZATION
25 DIR, USARL
AMSRL-WM-TC,
W S DE ROSSET
T W BJERKE
R COATES
FGRACE
K KIMSEY
M LAMPSON
D SCHEFFLER
S SCHRAML
G SILSBY
B SORENSEN
R SUMMERS
W WALTERS
AMSRL-WM-TD,
A M DIETRICH
D DANDEKAR
KFRANK
M RAFTENBERG
A RAJENDRAN
G RANDERS-PEHRSON, LLNL
M SCHEIDLER
S SCHOENFELD
S SEGLETES (2 CPS)
T WEERISOORIYA
AMSRL-WM-TE,
J POWELL
A PRAKASH
35
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
36
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
37
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
38
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
39
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
40
NO. OF NO. OF
COPIES ORGANIZATION COPIES ORGANIZATION
UK MINISTRY OF DEFENCE
G J CAMBRAY
CBDE PORTON DOWN SALISBURY
WITTSHIRE SPR OJQ
UNITED KINGDOM
K TSEMBELIS
SHOCK PHYSICS GROUP
CAVENDISH LABORATORY
PHYSICS & CHEMISTRY OF SOLIDS
UNIVERSITY OF CAMBRIDGE
CAMBRIDGE CB3 OHE
UNITED KINGDOM
UNIVERSITY OF KENT
PHYSICS LABORATORY
UNIT FOR SPACE SCIENCES
PGENTA
P RATCLIFF
CANTERBURY KENT CT2 7NR
UNITED KINGDOM
41
INTENTIONALLY LEFT BLANK
42
Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704-0188
Public reporting burden tor this collection ol Information Is estimated to average 1 hour per response. Including the tin» tor reviewing Instructions, searching existing data sourcesT
gathering and maintaining the data needed, and completing and reviewing the collection of Information. Send ccmrwitsregardngthla burden eatknäte or any other aspect ot this
colsetlon of Intormatlon, Including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1i16 Jefferson
Davis Hlnhwav. Sun«. 1204. «mnnton. VA VJOMSm. m«, la the Office ol M.n.««n<«l and Budnet. fWwon. Reduction Prol«cl(07044)188l. Washington. DC 20M3.
1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
February 1999 Final, Sep - Dec 98
4. TITLE AND SUBTITLE 5. FUNDING NUMBERS
43
Bernoulli, noninertial, streamlines, penetration, acceleration 16. PRICE CODE
17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT
UNCLASSIFIED UNCLASSIFIED UNCLASSIFIED UL
NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)
43 Prescribed by ANSI Std. 239-18 298-102
INTENTIONALLY LEFT BLANK.
44
USER EVALUATION SHEET/CHANGE OF ADDRESS
This Laboratory undertakes a continuing effort to improve the quality of the reports it publishes. Your comments/answers
to the items/questions below will aid us in our efforts.
1. ARL Report Number/Author ARL-TR-1895 (Sesletes'l Date of Report February 1999
2. Date Report Received
3. Does this report satisfy a need? (Comment on purpose, related project, or other area of interest for which the report will
be used.)
4. Specifically, how is the report being used? (Information source, design data, procedure, source of ideas, etc.).
5. Has the information in this report led to any quantitative savings as far as man-hours or dollars saved, operating costs
avoided, or efficiencies achieved, etc? If so, please elaborate.
6. General Comments. What do you mink should be changed to improve future reports? (Indicate changes to organization,
technical content, format, etc.)
Organization
7. If indicating a Change of Address or Address Correction, please provide the Current or Correct address above and the Old
or Incorrect address below.
Organization
OLD Name
ADDRESS
Street or P.O. Box No.
DIRECTOR
US ARMY RESEARCH LABORATORY
ATTN AMSRLWMTD
ABERDEEN PROVING GROUND MD 21005-5066